Hi,
the static analysis plugins hold their state in Maps. At least this has
been the state of affairs last tie i looked. Kohsuke supplied patches
that reduce this amount but when you have the count of warnings you
mentioned it is simply a problem of allocating memory to your
application server instance running jenkins.
At least as long as the storage backend of the analysis plugins is not
changed.
So the gist is: give your jenkins more memory and it won't commit
suicide anymore.
regards
Dirk
On 10.11.2012 08:33, Varghese Renny wrote:
> Check your build where you are mentioning ant task..take advanced option
> adjacent to it...there give jvm memory...
> as argument..
>
> On Fri, Nov 9, 2012 at 8:38 PM, David Weintraub <[email protected]
> <mailto:[email protected]>> wrote:
>
> There's not a lot of jobs running on this. We're just getting
> Jenkins setup, and maybe get one or two builds per day. The problem
> seems to stem from the Analytic plugins: Findbugs, PMD, Checkstyle,
> Warnings, and CPD. The build completes fine, but when it is
> calculating the issues from the previous build to the present, it
> runs out of memory (usually during PMD).
>
> I never used Tomcat before and was wondering if it was related to
> that because I never had a memory issue before. Last place where I
> worked, we had six build executors and Jenkins was setup to use only
> 512mb of heap memory. Then again, these analytics are finding tens
> of thousands of issues a piece. The last time PMD ran, it found
> almost 65,000 issues with our code. Same with Findbugs. Even Javadoc
> comes up with almost 1000 warnings. The Ant tasks run and complete,
> but the Analysis of these issues is where I get memory issues.
>
> Maybe our code is in such bad shape that during the analysis Jenkins
> gets depressed and commits suicide.
>
>> You have to set JVM in two places, one for system, one for the
>> particular job you are running..check it out.
>
> Never thought of that. Where do I set Jenkins per job? Can Jenkins
> spool a job separately? Or, is this only when it's running Ant
> (where I can set the memory on a per job basis). The problem is that
> this is a post-build task that's crapping out.
>
> For now, I've just eliminated the analysis. Maybe I'll turn them on
> one at a time once the developers decide to go back and clean up
> their code.
>
>
> On Nov 9, 2012, at 12:14 AM, Varghese Renny
> <[email protected] <mailto:[email protected]>> wrote:
>
>> Hi David,
>>
>> You have to set JVM in two places, one for system, one for the
>> particular job you are running..check it out..I think 2GB memory
>> is more than enough for one job..You can analze it through
>> monitoring plugin..
>>
>> Options are you can dump your heap memory to some location in your
>> system and use some tools to anlayse those dumped memory to find
>> any memory leakage..I think in eclipse one memory analyzer plugin
>> is there. You can search some other better tools also..
>>
>>
>> Regards,
>> varghese
>>
>>
>>
>> On Fri, Nov 9, 2012 at 2:59 AM, Qazwart <[email protected]
>> <mailto:[email protected]>> wrote:
>>
>> At first, I was getting Out of Heap errors and increased the
>> server's memory requirements to 1024m.
>>
>> Now I have a FATAL: GC overhead limit exceeded error.
>>
>> I'm running Jenkins 1.476 on a Redhat server running on Tomcat
>> 7.0.27.
>>
>> I have the following options set: -Xmx1024m
>> -Xx:PrintGCTimeStamp -verbose:gc -XX:-UseGCOverhradLimit
>>
>> (The GC stuff I just added)
>>
>> I can see a full GC constantly being called every second.
>> This happens after the build is complete and after I am
>> running the PMD plugin.
>>
>> Any advice? I've increased memory to 2Gb. This is the first
>> time I used Jenkins with Tomcat. I'm wondering if there's an
>> issue with Tomcat.
>>
>> --
>> David Weintraub
>> [email protected] <mailto:[email protected]>
>>
>>
>
>
